How CESMII’s i3X Is Ending Manufacturing’s API Chaos

Executive Key Points Why manufacturers cannot skip the three smart manufacturing imperatives on the path to AI readiness How graph relationships, not hierarchies, connect equipment failures to real business impact Why generative AI cannot compensate for semantic inconsistency in your manufacturing data How i3X eliminates API chaos across mixed-vendor environments with a free, open-source specification What “a brutal mirror” reveals about where your operation actually stands on digital transformation Get future Practitioners Unplugged episodes Join manufacturing executives getting new conversations and key insights in their inbox. Subscribe Now “If you get the model right, the applications emerge from the model.” Someone smarter than Jonathan Wise said that to him once. He has not forgotten it. The problem is that manufacturing has been doing the opposite for decades. Applications get deployed first, each one carrying its own implicit data model, locked inside its own platform, inaccessible to anything around it. The result is an industry drowning in data it cannot use, chasing AI initiatives that stall before they deliver, and wondering why the promise of smart manufacturing keeps slipping further out of reach. Jonathan Wise, Vice President of Technology and Chief Architect at CESMII, returned to Practitioners Unplugged for Episode 19 to explain exactly why this keeps happening, and what manufacturers and vendors can do about it right now. The Foundation Most Manufacturers Are Skipping CESMII has organized the path to a functional smart manufacturing operation into three imperatives. Jonathan was direct about why the sequence matters: “You can’t do i3X without having worked on the previous two.” Imperative one is modeling your data into shareable templates. Every system in your operation already has some version of a data model, but those models are inconsistent and non-portable. Jonathan compared it to class definitions in software development: “In a strongly typed software development environment, I can’t even run the code if any of my data doesn’t comply with the class that it claims to be of.” CESMII’s preferred format for this is an OPC UA node set, a common templating language any vendor can use to share information templates between systems. Imperative two adds context by connecting those component models through graph relationships. This is where most organizations stop short. A hierarchical model, like an ISA-95 asset hierarchy, captures the static layout of a plant. Graph relationships capture how things interact over time. As Jonathan explained: “Material has a relationship with the machine for some period of time, and then that relationship goes away. An operator has a relationship with the machine for some period of time, and then that relationship goes away or maybe changes to point to a different operator.” Dante put the business case plainly: “I drive predictive analytics in a motor. Okay, your machine’s gonna fail. But what’s the impact? That’s gonna impact this order number. That’s gonna impact my cost of quality. It’s gonna impact my shipping downstream.” Without graph relationships, that chain of impact is invisible. Why AI Cannot Fix a Broken Foundation This is the part most organizations do not want to hear. Generative AI cannot compensate for poor data infrastructure, no matter how capable the model. Jonathan explained why: “The generative AI that we’re using to generate text depends on relationships between tokens that were learned over a massive corpus of semantically consistent, structurally consistent words. We have no semantic or structural consistency, never mind across our industry, even within an enterprise.” Sree offered a sharp observation in response. The one place inside most enterprises where semantic consistency actually exists is finance and accounting. That consistency is top-down enforced, and it is precisely why AI tools can now work meaningfully against financial data and APIs. Manufacturing has no equivalent, and closing that gap is exactly what imperative three is designed to address. Imperative Three: i3X and the End of API Chaos After analyzing 50 smart manufacturing projects nationwide, CESMII identified eight to ten functions that any complete information system must be able to perform: exploring the information model, querying objects, accessing historical data, subscribing to changes, and writing back updates. Those functions became the foundation of i3X, an open-source API specification designed to eliminate what Jonathan calls “API chaos” across mixed-vendor environments. The scale of that chaos is real. A typical manufacturer runs 10 to 30 software packages, each with its own proprietary API, each requiring custom integration work to share data with anything else. Every upgrade, every platform transition, every new tool added to the stack breaks something. i3X addresses that problem by giving every platform a common compatibility layer, a single agreed-upon interface that any application can work against regardless of what sits underneath. The specification is fully open source, MIT licensed, and free. As Jonathan put it: “You don’t have to be a CESMII member. You don’t have to sign up for anything. You can go read it right now on GitHub and see the demo endpoint live and interact with it live, for zero dollars with no license, no requirements at all.” Adoption has moved faster than expected. After a community preview at the Prove It conference and a beta announcement at Hannover Messe, approximately 30 vendors had already committed to implementation. Most wrap an existing API in two weeks or less. A junior developer can work through the SDK in 20 to 30 minutes. AI-assisted setup takes around three minutes. A Brutal Mirror for Manufacturers and Vendors Alike Working group member Matt Paris described i3X with a phrase Jonathan clearly appreciated: “i3X is a brutal mirror.” For manufacturers who have already done the foundational work, i3X lights up quickly and validates the investment. For those who have not, it makes the gaps impossible to ignore. “You can’t pretend you’re done. You can’t pretend you’re Industry 4.0 or AI-ready. You don’t have the primitives necessary to complete that modernization.” Sree framed the longer arc for business leaders watching from a distance. The parallel to HTTP and HTML is instructive. Twenty years ago, business systems connected to users through a wide variety of proprietary protocols. Gradually, the industry agreed to eliminate that unnecessary variety, and an explosion of value followed. i3X is making the same bet for manufacturing data. For vendors, Jonathan’s call to action was unambiguous: stop selling integration complexity as a service. “Just wiring things up over and over and over again isn’t actually returning value to your customers. Working against a model to predict something, to optimize something, to help relieve a business constraint, there’s a ton of value to be had and harvested there.” Where to Start For end users, start by asking your vendors where they are on their i3X adoption journey. For vendors who have not yet engaged, the barriers are low and the working group is active. For those who want to shape the specification before it potentially moves to a formal standards body, the GitHub is open and all discussions are transparent. The i3X specification, SDK, and live demo endpoint are available now at i3x.dev. To learn more about CESMII’s broader knowledge programs, technology initiatives, and how to get involved in the collaborative ecosystem, visit CESMII.org. The promise of AI in manufacturing does not fail because the ambition is wrong. It fails because the foundation was never built. CESMII’s three imperatives, and i3X as their culmination, are the most credible framework available for manufacturers who are ready to stop talking about Industry 4.0 and actually build it. The tools are free. The community is open. The only thing left is the decision to start. Keep practicing. Keep learning. Keep transforming. To submit a request for a new episode topic from Practitioners Unplugged, visit our contact page Thanks for reading. Don’t forget to subscribe to our weekly newsletter to get every new episode, blog article, and content offer sent directly to your inbox. Explore more about Schneider Electric & AVEVA